Commits

Branch selector

 levit

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 6, 2021

• fix images

  lucidrains

  committed
  db04c0f

• complete levit

  lucidrains

  committed
  2cb6b35

• levit without pos emb

  lucidrains

  committed
  2ec9161

Commits on Apr 2, 2021

• add layer dropout for CaiT

  lucidrains

  committed
  3a3038c

Commits on Apr 1, 2021

• offer hard distillation as well

  lucidrains

  committed
  b1f1044

• readme

  lucidrains

  authored
  deb9620

• make sure layerscale epsilon is a function of depth

  lucidrains

  committed
  05b47cc

• add CaiT, new vision transformer out of facebook AI, complete with layerscale, talking heads, and cls -> patch cross attention

  lucidrains

  committed
  9ef8da4

Commits on Mar 31, 2021

• add documentation for three recent vision transformer follow-up papers

  lucidrains

  committed
  506fcf8

• add arxiv links for now, document in readme later

  lucidrains

  committed
  6fb360a

• cite

  lucidrains

  committed
  9332b9e

• add working PiT

  lucidrains

  committed
  da950e6

• use depthwise conv for CvT projections

  lucidrains

  committed
  4b9a02d

Commits on Mar 30, 2021

• add CvT

  lucidrains

  committed
  518924e

• add CrossViT

  lucidrains

  committed
  e712003

• make recorder work for t2t and deepvit

  lucidrains

  committed
  d04ce06

Commits on Mar 29, 2021

• use hooks to retrieve attention maps for user without modifying ViT

  lucidrains

  committed
  8135d70

• add recorder class, for recording attention across layers, for researchers

  lucidrains

  committed
  3067155

Commits on Mar 28, 2021

• cleanup

  lucidrains

  committed
  ab7315c

Commits on Mar 23, 2021

• remove masking, as it complicates with little benefit

  lucidrains

  committed
  15294c3

• add deep vit

  lucidrains

  committed
  b900850

Commits on Mar 10, 2021

• readme

  lucidrains

  authored
  7848904

Commits on Mar 8, 2021

• cleanup and release 0.8.0

  lucidrains

  committed
  173e07e

• Merge pull request lucidrains#66 from zankner/masked_patch_pred

  Show description for 0e63766
  lucidrains

  authored
  0e63766

• added to readme

  zankner

  committed
  a6cbda3

Commits on Mar 7, 2021

• converting bin targets to hard labels

  zankner

  committed
  73de1e8

Commits on Feb 26, 2021

• make it so one can plug performer into t2tvit

  lucidrains

  committed
  1698b7b

Commits on Feb 24, 2021

• no need to do projection to combine attention heads for T2Ts initial one-headed attention layers

  lucidrains

  committed
  6760d55

Commits on Feb 22, 2021

• add DistillableT2TViT

  lucidrains

  committed
  a828948

• remove patch size from T2TViT

  lucidrains

  committed
  3744ac6

• make sure distillation still works

  lucidrains

  committed
  6af7bbc

Commits on Feb 20, 2021

• cleanup

  lucidrains

  committed
  05edfff

• add token to token ViT

  lucidrains

  committed
  e3205c0

Commits on Feb 17, 2021

• incept idea for using nystromformer

  lucidrains

  committed
  4fc7365

• fix for ambiguity in broadcasting mask

  lucidrains

  committed
  3f2cbc6

Pagination

PreviousNext